Modeling of an industrial wet grinding operation using data-driven techniques

Real time optimization based on a serial hybrid model for gold cyanidation leaching process

“…The former corresponds to that the first-principle model is the basis, whereas the data-driven model is used to predict the deviation between the first-principle model output and the real output (Lee et al, 2005;Mitra and Ghivari, 2006). The latter is an alternative modeling method that the known parts are modeled by the first-principle model, whereas the unknown parameters are predicted by the data-driven models (Mjalli and Ibrehem, 2011;Gupta et al, 1999;Psichogios and Ungar, 1992).…”

Optimization of time‐variable‐parameter model for data‐based soft sensor of industrial debutanizer

“…However, there is another class under the name of hybrid (gray box) which is a combination of model‐driven and data‐based methods. It is noted that the search for the optimal model structure should be approached more systematically, and therefore, the development of identification procedures and data‐based mechanistic (DBM) has received significant attention from different modelers . The statistical methods or soft computing have been applied in different data‐based soft sensors.…”

Particle size estimate of grinding processes using random vector functional link networks with improved robustness

“…Based on a training data set containing representative behavior of the grinding process, various learning algorithms (e.g. case-based reasoning [7], fuzzy logic [8], support vector regression, artificial neural network [9], and genetic algorithms [10]) were employed to explore the mapping of the process variables to the PS. But these data-driven models are of pure black box structure, which results in difficulty in finding the optimal model structure and parameters.…”